- Email: [email protected]
Heavy gold cluster beams production and identification
Nuclear Instruments and Methods in Physics Research B 225 (2004) 579–589 www.elsevier.com/locate/nimb
Heavy gold cluster beams production and identification S. Bouneau, S. Della-Negra, J. Depauw, D. Jacquet, Y. Le Beyec, J.P. Mouffron, A. Novikov, M. Pautrat * Institut de Physique Nucleaire, IN2 P3 -CNRS, Av. Georges Clemenceau, 91406 Orsay, France Received 4 March 2004; received in revised form 25 May 2004
Abstract It is shown that beams of very heavy gold clusters can be produced by a liquid metal ion source (LMIS). An experimental method is described for defining the LMIS source and the Wien filter parameters that must be set to extract and select large Aun clusters. This method is based on the acceleration of the clusters to high energy (MeV) and on the measurement, after their passage through a thin foil, of their number of constituents and velocity. Only an average mass over charge value is obtained for a given set of source and Wien filter parameters. These parameters can then be used to select heavy Aun cluster beams for applications at low energy (keV) in mass spectrometry. 2004 Elsevier B.V. All rights reserved. PACS: 36.40.-c; 36.40.Wa; 36.10.+j Keywords: LMIS; Heavy gold clusters; Mass and charge identification; Multipixel detector
1. Introduction Beams of gold clusters, produced by a liquid metal ion source (LMIS), have been used since the early nineties to study collisions with various types of materials [1–6]. A large range of projectile energies was investigated, mainly in experiments on sputtering phenomena and secondary emission processes. The highest energies of Aun clusters ðn ¼ 1; 13Þ were obtained by using heavy ion accelerators at Orsay and Lyon. During the course
*
Corresponding author. Tel.: 33-1-69157323; fax: 33-169154507. E-mail address: [email protected] (M. Pautrat).
of these experiments it was observed that large size clusters could be produced with the gold LMIS and it became necessary, for further experiments, to know the size (i.e. the number of constituents) and the total electric charge q of these entities delivered by the source. One of the Aun sources, including the ion optics and a Wien filter, is located at the Orsay MP Tandem terminal. A wide mass range of clusters corresponding to nq ratios of 5, 7, 9, 27, 40 (selected by the Wien filter) and even nanodroplets containing up to 400 atoms are produced, accelerated by the second stage of the accelerator and sorted out by the analysing magnet. According to the parameters set on the source and the Wien filter the selection of nq values is obtained with a distribution of charge and mass.
0168-583X/$ - see front matter 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.nimb.2004.06.005
580
S. Bouneau et al. / Nucl. Instr. and Meth. in Phys. Res. B 225 (2004) 579–589
In electrostatic accelerators the total particle energy depends on its charge. Therefore the measurements, after acceleration, of the cluster constituent numbers and velocities allow identification of the clusters (for a given tuning of the source). This is achieved by time-of flight measurements of the cluster atomic fragments, after passage of the incident beam through a thin Formvar foil and by measuring simultaneously, with a multianode detector [7], the number of constituents coming out of the foil. High energy clusters, accelerated at several MeV, are thus necessary to define and verify experimentally the working parameters of the cluster ion source device. The same type of source is now used to produce large size clusters with energies below 50 keV to investigate desorption phenomena with large projectiles as well as applications in mass spectrometry. For example, the secondary ion emission from bio-organic solids under the impact of a Au4þ 400 beam has been studied for the first time [8]. Several sets of source parameters have been experimentally explored and this work presents the method to characterize Aun cluster beams and the variety of beams which can be delivered by a gold LMIS.
2. Experimental process
selection of the ejected clusters is located at the accelerator terminal [10,13]. The first cluster beams were obtained, over a large energy range, more than ten years ago. To increase the cluster beam intensities injected into the accelerator, the diameter of the extractor aperture at the source exit is set at 1.5 mm which leads to a 75 mrad acceptance. m In this configuration, the mass separation Dm is between 15 and 20. With this device it is possible to obtain ions from Siþ to Auþ 1000 . Fig. 1 shows an example of mass spectrum for very heavy ions down to Auþ 5 . It shows the variation of the intensity measured by the Faraday cup, at the exit of the Wien filter, versus the voltage set on the filter plates. Clusters containing some tens of atoms were already obtained with this source [14,15], but in this spectrum a maximum of beam intensity is observed around a potential value of 10 V corresponding to mq 20 000. This ratio is deduced from the relation rffiffiffiffi q VWien þ d ¼ K I ; m VWien (V): voltage on the electrostatic deviation plates, I (A): current intensity in the coils generating the magnetic field, K: constant depending on the alignement of the source and Wien filter axes, experimentally adjusted.
2.1. Beam production 2.1.1. The ORION injector The gold clusters are provided by a liquid metal ion source, duplicate of the pattern developed by Sudraud and Ben Assayag [9] the specifications of which are described in [2,10,11]. This simple source is made of a tungsten needle and of a reservoir containing the metal. An electric field, applied between the needlepoint and an extractor, generates a Taylor cone from which are emitted ions, clusters and nanodroplets. These are further accelerated by a potential of some tens of kilovolts. To obtain a low melting point, 370 C, the reservoir contains an AuSi eutectic with 31 silicon atoms for 100 gold ones (i.e. a mass ratio of 4.4% silicon with respect to gold) [12]. The whole set-up including the ion source, two focusing lenses and a Wien filter for the mass
Fig. 1. Current delivered by the LMIS source versus the VWien voltage.
S. Bouneau et al. / Nucl. Instr. and Meth. in Phys. Res. B 225 (2004) 579–589
The offset value d depends on the beam angle at the filter inlet and a mass calibration of the Wien filter, with well known beams (Auþþ , Auþ , Auþ 2, Auþ 5 , . . .), is advisable for each new source configuration. Above nq ¼ 9 it is not possible any more to separate the cluster masses since the curve of Fig. 1 becomes continuous. However, one can modify the intensity of the cluster beam as well as the maximum of its mass distribution through the adjustment of three main parameters: the voltages on the focusing lenses, the current emitted by the source and the acceleration voltage.
581
i(ii) Fig. 3 shows the importance of the source emission current, as large size cluster beams, obtained for low values of the Wien filter potential (Fig. 1), have their highest intensity for I ¼ 100 lA. But a compromise must be found between the current value and the lifetime of the ion source (from 300 to 1000 h). (iii) The acceleration voltage value is, in this case, fixed at 10 kV. In the following, since the same nq ratio may be related to different cluster size and charge state values, two sets of source parameters are used for producing either low charged or highly charged clusters, which also corresponds to light or heavier ones.
ii(i) The voltages (U1 and U2 ) applied on the two focusing lenses allow to optimize the intensity of the selected clusters, compensating for the energy spread due to the ionization of clusters at some distance from the needle head. This energy spread increases with the cluster mass [15] and can reach 100 eV or more. As can be seen in Fig. 2, the focusing conditions play an important role for ‘‘light’’ clusters but are nearly the same from nq ¼ 9 up to 100. However, the focusing conditions can also change the shape of the beam intensity curve. n The Auþ n peaks, for q P 9, may disappear while a ‘‘background’’ increases which is not an actual background but the sum of the 3þ 2þ 3þ Au2þ 2n1 , Au3n1 , Au2nþ1 , Au3nþ1 ,. . . peaks.
2.1.2. Acceleration and beam line At the Tandem terminal the selected clusters are injected, using two electrostatic mirrors, in the high energy (HE) accelerating tube in order to reach energies from 1 to 12q MeV. The residual gas pressure in the tube and along the beam line is kept lower than 107 h Pa by ion and titanium vapor pumps. The beam line includes, at the machine exit, two electrostatic quadrupole lenses together with X and Y steerers, to focus and direct the beam to an analysing magnet followed by a diagnostic chamber. In the latter, by means of sliding slits associated with a microchannel plate
Fig. 2. Optimized values of the voltages (U1 , U2 ) applied on the two focusing lenses versus the n=q ratios of the selected clusters.
Fig. 3. Intensity of the cluster beam as a function of the Wien filter potential for different source emission currents.
582
S. Bouneau et al. / Nucl. Instr. and Meth. in Phys. Res. B 225 (2004) 579–589
detector (MCP) and a Faraday cup, the beam can be centered and collimated. For nq 6 40, the diagnostic and collision chambers are set at a 1.29 angle with respect to the beam axis and the analysing magnet eliminates the fragments of the projectile, mainly due to collisions with the residual gas. The magnetic field necessary to bend the heavier cluster trajectories would be too high and the measurements are then performed at 0 without rejection of the background due to the fragments.
detector records, for each incident projectile, the number of simultaneous impacts (multiplicity) on the detector. The data accumulation over a large number of events provides a multiplicity distribution with a centroid M. Knowing the detector efficiency (see Appendix A), the average number of constituents n ¼ M is obtained as well as the average charge q as nq ¼ nq. Furthermore, the time-of-flight measurement between the foil and the multianode detector gives the velocity of the Au atoms exiting the Formvar foil:
2.2. Method used to characterize the ion beams DT ¼ A schematic view of the experimental set-up is given in Fig. 4. The clusters with given nq values are selected by the Wien filter, pre-accelerated by a dU voltage in the LMIS source device and further accelerated from the terminal at a potential U to the collision chamber. The total energy for a charge q cluster is then E ¼ q½Uterminal þ dULMIS :
ð1Þ
For each charge q there is a distribution of n due to the limited mass separation power of the Wien filter. A variety of clusters, with different q values and constituent numbers, are thus accelerated. After passing through a Formvar foil perpendicular to the beam direction, a cluster dissociates into its atomic constituents and the pixelated
L L ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffi ; 2ðE1 DEÞ v mAu
where DE is the calculated energy loss in the thin foil, mAu the gold atomic mass and L the foildetector distance. The time-of-flight measurement of the atomic constituents allows verification of the correct assignment of the nq ratio, deduced from the Wien filter parameters. Experimentally a window set on the time-of-flight peak is then used to remove the background in the measured distribution of the constituent number. 2.3. Detection 2.3.1. Short description The electrons and negative ions emitted at the foil entrance are deflected at 90 by an electrostatic
Fig. 4. Schematic view of the experimental set-up.
S. Bouneau et al. / Nucl. Instr. and Meth. in Phys. Res. B 225 (2004) 579–589
mirror towards an MCP detector; the pulse delivered by the electron impact gives the start of the time-of-flight (Fig. 4). Furthermore, for nq P 40, the shape of the signals due to the electrons is visualized and recorded by a digital oscilloscope, to make sure that they actually correspond to a single impact and not to several successive fragments. It is also possible to choose the H ions as a start signal. The projectile constituents, after the foil, are collected by a multianode detector; two different detectors of this kind have been used. The first one, described in [7], is made of a set of two MCPs (Hamamatsu, / ¼ 42 mm, geometrical efficiency 0.6) and of a 256 pixel anode array (1600 mm2 ); it will be called D256 hereafter. The other smaller one (D64), has three MCPs (Hamamatsu, / ¼ 28 mm, geometrical efficiency 0.66) and a 64 pixel anode array (400 mm2 ). With these detectors, located along the beam axis at 256 mm from the target, it is possible to measure the number of constituents and their time-of-flight. The use of the SRIM code [16] to compute the angular distributions of the constituents coming out of the foil allows calculation of the proportion reaching the detector, listed in the last column of Table 1. Four formvar foils, with known thicknesses, are used in these experiments and the time-of-flight Table 1 Summary of experimental conditions n Energy/Au Deviation q (MeV) angle ()
583
calibration is performed with a pure Au5 beam at a known energy. 2.3.2. Detector efficiency A more complete treatment of this subject is reported in Appendix A. Two efficiencies are involved for the detector: 1 is obtained by multiplying, for each beam, the MCPs geometrical efficiency by the percentage of intercepted ions and 2 takes into account the pixelated structure of the anode array. The efficiency 2 is related to the geometrical structure of the anode and to the discriminator thresholds [7]. The current delivered by the set of MCPs, in these experiments, is always sufficient to detect the impact of gold atomic ions. The 2 value corresponds to the detection of an impact giving a single triggering in the pixels, whereas (1 2 ) is associated to the detection of an impact in the interpixel zone leading, in our case, to a simultaneous triggering of two adjacent pixels. The 2 value is close to the ratio 2 ’
active geometrical surface ¼ 0:74: total geometrical surface
We took this value as a starting point for the multiplicity analysis, given in detail in the next section, and 2 has been adjusted to get the best agreement with experimental results.
Formvar foil thickness (nm)
Detector
Ion source conditions
% of intercepted ions
5 7 9 27
2.008 1.434 1.115 0.372
1.29 – – –
49 ± 2 – – –
D64 / ¼ 28 mm – –
Low charge Mode – –
66 53 44 13
5 40
1.808 0.225
1.29 –
25 ± 2 –
D256 / ¼ 42 mm
Low charge Mode
100 58
35 120 400
0.287 0.084 0.025
0 0.03 0
15 ± 2 – –
D256 – –
Low charge Mode –
81 38 22
9 27 40
0.2 0.2 0.2
1.29 – –
26 ± 2 – –
D256 – –
High charge Mode –
40 – –
584
S. Bouneau et al. / Nucl. Instr. and Meth. in Phys. Res. B 225 (2004) 579–589
3. Experimental results and discussion The LMIS source delivers Aun Sip clusters with a p=n ratio of 0.1 to 0.2, estimated from the mass spectra. In this paper, we take into account the number of gold constituents of the cluster, neglecting the silicon contribution. Through this approximation only 2% of the cluster mass is, in fact, neglected, which has no visible effect on the time-of-flight spectra. The silicon atoms are scattered, after the foil, in a solid angle about five times wider than that for the gold atoms and less than 10% of these reach the detector. So, the number of gold constituents measured is overestimated by 2% at the most, because of the unknown silicon atoms number. The experimental conditions for the various beams are summarized in Table 1. Two sets of experiments are performed and compared: one for a given total energy of the projectiles and the other for the same projectile velocity. 3.1. Time-of-flight For a given energy per atom (the total energy and nq ratio being known) one can calculate the energy loss in the Formvar foil and deduce the time-of-flight values. The measured time-of-flights values, up to nq 400, agree with the calculated ones. Furthermore, the energy spread in the foil, as well as scattering phenomena, give rise to angular
and energy distributions of the exiting atoms which can be estimated with the SRIM code [16]. As a result, the calculated time-of-flight peak shapes are also shown to be in agreement with the experimental data. An example of calculated curve is given in Fig. þ 5 for the Auþ 5 and Au9 projectiles, at the same 10.04 MeV total energy. The adjustment of these curves, as well as that of the Auþ 7 one (not shown), is obtained through the convolution of the time-offlight distribution, deduced from the SRIM energy distribution of the atomic ions, with a Gaussian distribution (r ¼ 3:8 ns) which accounts for the time-of-flight resolving power and the nonhomogeneity of the Formvar foil. The parameters used in the three cases are the same. 3.2. Determination of the gold cluster masses Fig. 6 shows the multiplicity (i.e. the number of impacts) distributions measured, in coincidence with a time window set on the time of flight peak, without magnetic selection, for voltages on the Wien filter plates corresponding to nq ¼ 35, 120, 400. In Fig. 6, the mean and maximum values of the measured distribution of impact numbers increase with the nq selection of the Wien filter and it is possible to obtain a rough nq estimate. A peak of high intensity at low multiplicity, due to the previously mentioned fragmentations, disappears when the beam is selected by the H.E.
7000 +
Au5
1000
5000
Counts/channel
Counts/channel
6000
4000 3000 2000 1000
180
(a)
+
Au9
100
10
230 240 250 260 270 280 290 300 310 320 330
200
Time (ns)
fit Experimental data
(b)
Time (ns)
þ Fig. 5. Experimental and calculated time-of-flight spectra for the Auþ 5 (histogram ¼ experiment) and Au9 projectiles (the time-of-flight offset value is 89 ns in both cases).
S. Bouneau et al. / Nucl. Instr. and Meth. in Phys. Res. B 225 (2004) 579–589
585
6
10
5
n/q=120
4
10 Counts
Probability of detection
10
n/q=35 3
10
2
10
1
10
n/q=400
0.1
0.01 M1 M2 M3 M4 M5 Sum of all contributions experimental data
1E-3
0
10
0
100
200
300
400
500
Number of impacts Ni on the D256 multi-pixel detector
Fig. 6. Number of counts versus the measured multiplicity, without magnetic field and for several Wien filter settings, in the low mass ion source conditions.
analysing magnet. However, for nq P 40 it is possible to use the H signal, which emission yield exceeds 100%, as a start signal to reduce the contribution of the fragmentation in the spectra. The emission of H ions induced by fragments is, indeed, nearly negligible. 3.2.1. Influence of the multianode structure There are two stages for the calculation of the number of constituents corresponding to the detection of the clusters. The probability for the MCPs to detect p ions among n follows a binomial law function of 1 : P ðpÞ ¼ Cnp p1 ð1 1 Þ
np
:
The pixelated structure of the anode surface implies to take into account the 2 efficiency as defined before. The detection probability of all the anode central parts obeys also a binomial law, function of 2 , and the probability of detecting r ions among n is the product of these two binomials: P ðrÞ ¼ Cnr r1 r2 ð1 1 2 Þnr : An ion falling in the interpixel area may be detected by the two adjoining pixels. The binomial law is the same, but the corresponding multiplicity is M ¼ 2ðp rÞ. The variation of the detection probability in terms of the multiplicity finally assigns to each P ðrÞ value a M ¼ r þ 2ðp rÞ ¼ 2p r multiplic-
0
2
4
6
8
10
Multiplicity Fig. 7. Calculated contributions of the M ¼ 1–5 detected ions and comparison of their sum with the experimental distribution for Au5 .
ity. Then, the distribution of the detected ions as a function of the multiplicity reads, as before, XX P ðMÞ ¼ Cnp Cpr p1 ð1 1 Þnp r2 ð1 2 Þpr ; p
r
with M ¼ 2p r. The experimental multiplicity, obtained with the D256 detector, for a 9 MeV Auþ 5 beam, is displayed in Fig. 7. The measured multiplicity values go beyond 5 because of the multiple triggering of ions falling in the interpixel zone. Taking this into account as described above, an interpixel surface leading to 18% of double hits gives a perfect fit of the measured multiplicity spectrum. The events associated to one to five impacts (marked Mi , i ¼ 1–5) are detailed in Fig. 7. 3.2.2. Average mass measurement The various multiplicity curves, for the singlecharged clusters Au5 , Au7 , Au9 , were fitted with the already determined 1 efficiency (detection yield and solid angle calculation) and by adjusting the 2 parameter. An 2 value, lower than 0.74, indicates that the surface responsible for multiplicities of 2 is larger than the interpixel surface, this effect being related to the detector gain. For clusters with a nq ratio of 27 and 40, higher charge states must be considered to fit the distribution.
586
S. Bouneau et al. / Nucl. Instr. and Meth. in Phys. Res. B 225 (2004) 579–589
Table 2 Efficiencies of the two detectors for the studied clusters and measured average multiplicities M Source parameters
n=q
Low charge
5 7 9 27 40 120
High charge
D64
D256
E/atom (MeV)
1
2
M
E/atom (MeV)
1
2
2 1.43 1.11 0.37
0.42 0.28 0.24 0.08
0.57 0.6 0.73 0.7
3.46 3.4 3.38 9.04
1.8
0.60
0.82
0.225 0.084
0.35 0.23
0.8 0.8
0.2 0.2 0.2
0.24 – –
0.75 – –
9 27 40
In Table 2 are listed, for the different nq and energy values, the 1 and 2 efficiencies used and the measured average multiplicity; 1 and 2 have been adjusted, around the values provided by the experimental conditions, to reach the best agreement between measured and calculated distributions in terms of the multiplicity. The slight variation of 2 with the constituent number should be connected to the detector gain and to the influence of the velocity on the MCPs response under the impact of gold ions. 3.2.3. Charge state determination Fig. 8 spectra are obtained with the ion source parameters which favour the low mass cluster beams. From Fig. 8(a) one can see that, even in the
M 3.5
57.3 154 4.62 30.4 54.7
¼ 9 case, a small proportion of Auþþ 18 added to the prevailing Auþ contribution is necessary to repro9 duce the multiplicity distribution. In Fig. 8(b) are compared the experimental multiplicity distribution for nq ¼ 40 and the sum of the computed contributions of the different charge states involved. A good fit of the experimental result needs to include the q ¼ 1þ to 5þ charge states. The contributions of the various charge states which, when summing them up, give the best agreement between experiment and calculation for different nq values, are listed in Table 3, for two sets of the ion source parameters. The Table 3 values (percentages and deduced mean q value) are accurate within 10%. Table 3 shows that the relative contribution of the high charge states increases with nq, and that the
n q
Fig. 8. Comparison between the multiplicity distributions measured and computed for (a) nq ¼ 9 and (b) nq ¼ 40 taking into account several charge states.
S. Bouneau et al. / Nucl. Instr. and Meth. in Phys. Res. B 225 (2004) 579–589
587
Table 3 Percentages of the charge states involved for several nq values and for two ion source regimes n=q Low charge
High charge
q
q
1
2
5 7 9 27 40 120
98.7 94 96.6 24.6 1 3.1
1.3 6 3.4 18.8 11 7.4
3
4
19.2 36 16
15.5 44 20.9
13.8 7 18
13.4
9 27 40
23 4.1 2
74 13.3 4.5
3 26 12
34.8 25
16.6 31
6 19.5
ion source conditions play a crucial part. Going from the ‘‘low charge’’ to the ‘‘high charge’’ conditions, the average charge goes up by 80% for the ions with a nq ¼ 9 ratio, and by 35% for nq ¼ 40.
5
6
7
8
9
8.1 10.3
6
4.7
6
1.01 1.06 1.03 3 3.4 4.8 1.8 3.6 4.6
should now be available with a nanoparticle beam. Moreover, in the near future, the use of microbeams of these heavy clusters may be considered for imaging of biological samples, with 1–10 lm dimensions and on a 10 nm depth.
4. Conclusion Appendix A The LMIS source delivers gold clusters from qþ Auþ 2 to Au1000 or more. Previous results have shown that a cluster distribution exists for nq ratios between 10 and 1000 (i.e. masses between 2000 and 200 000 amu) with a maximum for nq ¼ 100 and with a beam intensity identical to that of Auþ 5. The cluster ions produced by a LMIS source and selected by a Wien filter are identified for a large qþ mass to charge range going from Auþ 5 to Au120q . The mass is deduced from the measurement of the constituent number of the cluster, accelerated at an energy of several MV q and broken up by passing through a thin foil. It is thus possible to obtain the mass and charge distributions of the cluster ions. For the maximum nq value explored here (120), the mean charge value is 5 and the mass distribution 9þ extends from Auþ 120 to Au1080 . A correlation has been established between the mass and charge distributions and the ion source working conditions. These experiments lead to a better knowledge of the LMIS source parameters and possibilities, in the fields of secondary emission and ion implantation in the keV energy range. Secondary ion emission rates greater than those ever observed
A.1. Featuring of the multianode detectors It is achieved following the procedure worked out in [7]. Three parameters must be determined: • The geometrical efficiency of the first MCP detector. • The detection loss due to the discriminator thresholds depending on the charge collected by the anodes, i.e. on the number of electrons emitted under ion impact and on the whole detector gain. • The efficiency of the anode array made of pixels with, in the interpixel zone, a separating grid surrounded by an insulating area to eliminate cross-talk between anodes. This configuration induces a special response of the interpixel. If the electron cloud coming out of the last MCP is in the interpixel zone, the respective positions of this cloud and of the interpixel center give rise to three possibilities: A single anode is hit and the incident ion is detected by one anode.
588
S. Bouneau et al. / Nucl. Instr. and Meth. in Phys. Res. B 225 (2004) 579–589
The two anodes are hit simultaneously and the ion is recorded twice. The charge collected by the anodes is not sufficient and the impact is not observed. The anode array efficiency splits up into the three probabilities:
Pdet : ¼
Ncoinc: N1B 1 ; N0 N0B D
ðA:6Þ
1 ¼ Pdet : D:
P0 no detection, P1 detection by a single anode, P2 detection by two adjoining anodes,
ðA:7Þ
If a transmitting grid T is on the beam path, Eqs. (A.6) and (A.7) become
with the relation P0 þ P1 þ P2 ¼ 1: ðA:1Þ The basic idea for the determination of the absolute efficiencies, or detection probabilities, is the coincidence between two detectors recording the same event. If the total number of particles is NTot: the first detector gives NRef: ¼ N0 ¼ NTot: :
Actually, account must be taken of the detector backgrounds, N0B and N1B , as well as of the correction for the data loss in the discriminators (D). The detection probability Pdet : is then written as
ðA:2Þ
Pdet : ¼
Ncoinc: N1B 1 ; N0 N0B DT
ðA:8Þ
1 ¼ Pdet : DT :
ðA:9Þ
If the detection solid angle does not cover the whole emission angle, it is necessary to add another term x describing the average collection percentage of the detector This term is equivalent to the transmission of a grid set on the incident ions path,
For the second detector, which must be calibrated,
1 ¼ Pdet : DT x:
N1 ¼ 1 NTot:
ðA:3Þ
In our detector case, with a pixelated anode array, two procedures are available:
Ncoinc: ¼ 1 NTot:
ðA:4Þ
Ncoinc: N0
ðA:5Þ
• either a small dimension beam is used to scan the detector, which provides its actual efficiency in each ðx; yÞ point (the coincidence number then corresponds to an event with at least one hit anode),
and the number of coincidences is
1 ¼
is the effective detection probability.
ðA:10Þ
(a) (b)
0.6
0.1
0.4 Pixel 3. Multiplicity 1 Pixel 4. multiplicity 1 Pixel 3. Multiplicity 2 Pixel 4. Multiplicity 2 Total
0.3 0.2
Efficiency
Efficiency
0.5
Total Pixel 3. multiplicity 1 Pixel 4. multiplicity 1 Pixel 3. multiplicity 2 Pixel 4. multiplicity 2
0.01
0.1 0.0 1.2
1E-3
1.4
1.6
Position X (mm)
1.8
10.0
15.0
20.0
30.0
40.0
50.0
60
Threshold (mV)
Fig. 9. D64 efficiency measured (a) versus the detector position, for a 30 mV threshold (b) versus the threshold in the middle of the interpixel zone.
S. Bouneau et al. / Nucl. Instr. and Meth. in Phys. Res. B 225 (2004) 579–589
1 ¼
1 S
Z Z ðx; yÞ dx dy;
ðA:11Þ
S
where S is the surface of the detector or of an anode array cell; • or the beam covers the whole detector and the effective efficiency takes into account the anode efficiency as for a grid crossing the ions path (Eqs. (A.8) and (A.9)). To find out Eq. (A.1) terms, a scan with a sharp beam tests the interpixel response. The yields associated to a single anode or to two neighbouring ones are measured in terms of the position, and the P1 and P2 probabilities of Eq. (A.1) are so determined, as well as P0 marking the lack of detection. Fig. 9(a) shows an example of these measurements; it presents the response of an area of the D64 detector to a 100 lm · 100 lm Auþ 5 beam, with a 30 mV discriminator threshold. It can be seen, on this figure, that in the interpixel center (1.4 mm position) the multiplicity 2 yield is maximum (two active anodes), greater than the value for a single hit anode. On the other hand, the total detection yield is 26% smaller, in this position, than that obtained in the middle of the adjoining pixel. We so have indeed access to P0 ðx; yÞ, P1 ðx; yÞ, P2 ðx; yÞ. Obviously, the P0 , P1 and P2 probabilities do vary with the charge per impact delivered by the detector, and/or with the discriminator threshold as shown in Fig. 9(b) where this analysis is made in the middle of the interpixel. In our case, we take into account P1 and P2 , while P0 is included in 1 . The notations used in this work are 2 ¼ P1
and
1 2 ¼ P2 :
For this study, the discriminator thresholds are set at 10 mV, so that the impacts producing two electrons at least are detected, and the detector gains are adjusted to obtain pulses with an average amplitude of about a hundred mV. The electron emission from a carbon foil following a gold atom impact is of about 10–20 electrons, with a factor of 2 between 300 keV and 2 MeV [18], and a full cluster impact follows a sublinear law in n0:6 , n being the constituent number [17,18]. With these conditions for the electron emission, it is possible to
589
neglect the 10 mV threshold as it has no effect. We were able to verify that D ’ 1 and that the interpixel loss is minimum, which agrees with P0 ’ 0.
References [1] M.G. Blain, S. Della-Negra, H. Joret, Y. Le Beyec, E.A. Schweikert, Phys. Rev. Lett. 63 (1989) 1625. [2] M. Benguerba, A. Brunelle, S. Della-Negra, J. Depauw, M. Joret, Y. Le Beyec, M.G. Blain, E.A. Schweikert, G. Ben Assayag, P. Sudraud, Nucl. Instr. and Meth. B 62 (1991) 8. [3] H.H. Andersen, A. Brunelle, S. Della-Negra, J. Depauw, D. Jacquet, Y. Le Beyec, J. Chaumont, H. Bernas, Phys. Rev. Lett. 80 (1998) 5433. [4] A. Brunelle, S. Della-Negra, J. Depauw, D. Jacquet, Y. Le Beyec, M. Pautrat, K. Baudin, H.H. Andersen, Phys. Rev. A 63 (2001) 022902. [5] S. Bouneau, A. Brunelle, S. Della-Negra, J. Depauw, D. Jacquet, Y. Le Beyec, M. Pautrat, M. Fallavier, J.C. Poizat, H.H. Andersen, Phys. Rev. B 65 (2002) 144106. [6] M. Fallavier, R. Kirsch, S.N. Morozov, J.C. Poizat, J.P. Thomas, N. Wehbe, Phys. Rev. B 68 (2003) 140102. [7] S. Bouneau, P. Cohen, S. Della-Negra, D. Jacquet, Y. Le Beyec, J. Le Bris, M. Pautrat, R. Sellem, Rev. Sci. Instr. 74 (2003) 57. [8] A. Tempez, J.A. Schultz, S. Della-Negra, J. Depauw, D. Jacquet, A. Novikov, Y. Le Beyec, M. Pautrat, M. Caroff, M. Ugarov, H. Bensaoula, M. Gonin, K. Fuhrer, A. Woods, Rapid Commun. Mass Spectrom. 18 (2004) 371. [9] G. Ben Assayag, Thesis, Universite Paul Sabatier, Toulouse, 1984. [10] P. Attal, S. Della-Negra, D. Gardes, J.D. Larson, Y. Le Beyec, R. Vienet-Legue, B. Waast, Nucl. Instr. and Meth. A 328 (1993) 293. [11] K. Boussofiane-Baudin, Thesis, Universite de Paris Sud, Orsay, 1993, IPNO-T-93-06. [12] Di Capua, Rend. Accad. Nazion. Lincei 29 (1920) 111. [13] B. Waast, S. Della-Negra, A. Lafoux, Nucl. Instr. and Meth. A 382 (1996) 348. [14] P. Sudraud, C. Colliex, J. van de Walle, J. Phys. Lett. 40 (1979) L207. [15] P.D. Prewett, G.L.R. Mair, in: P.S. Walsh (Ed.), Ion Beams from Liquid Metal Ion sources, Research Studies Press, 1991. [16] J.F. Ziegler, J.P. Biersack, U. Littmark, SRIM 2000.39, www.SRIM.org, in: J.F. Ziegler (Ed.), Stopping Powers and Ranges of Ions in Matter, Pergamon Press, New York, 1985. [17] M. Fallavier, Y. Champelovier, M. Ferrari, R. Kirsch, J.C. Poizat, J.P. Thomas, B. Canut, M. Monchanin, S.M.M. Ramos, P. Thevenard, S. Della-Negra, J.P. Mouffron, P. Nicol, Eur. Phys. J. D 9 (1999) 629. [18] K. Baudin, A. Brunelle, S. Della-Negra, J. Depauw, Y. Le Beyec, E.S. Parilis, Nucl. Instr. and Meth. B 117 (1996) 47.
Heavy gold cluster beams production and identification S. Bouneau, S. Della-Negra, J. Depauw, D. Jacquet, Y. Le Beyec, J.P. Mouffron, A. Novikov, M. Pautrat * Institut de Physique Nucleaire, IN2 P3 -CNRS, Av. Georges Clemenceau, 91406 Orsay, France Received 4 March 2004; received in revised form 25 May 2004
Abstract It is shown that beams of very heavy gold clusters can be produced by a liquid metal ion source (LMIS). An experimental method is described for defining the LMIS source and the Wien filter parameters that must be set to extract and select large Aun clusters. This method is based on the acceleration of the clusters to high energy (MeV) and on the measurement, after their passage through a thin foil, of their number of constituents and velocity. Only an average mass over charge value is obtained for a given set of source and Wien filter parameters. These parameters can then be used to select heavy Aun cluster beams for applications at low energy (keV) in mass spectrometry. 2004 Elsevier B.V. All rights reserved. PACS: 36.40.-c; 36.40.Wa; 36.10.+j Keywords: LMIS; Heavy gold clusters; Mass and charge identification; Multipixel detector
1. Introduction Beams of gold clusters, produced by a liquid metal ion source (LMIS), have been used since the early nineties to study collisions with various types of materials [1–6]. A large range of projectile energies was investigated, mainly in experiments on sputtering phenomena and secondary emission processes. The highest energies of Aun clusters ðn ¼ 1; 13Þ were obtained by using heavy ion accelerators at Orsay and Lyon. During the course
*
Corresponding author. Tel.: 33-1-69157323; fax: 33-169154507. E-mail address: [email protected] (M. Pautrat).
of these experiments it was observed that large size clusters could be produced with the gold LMIS and it became necessary, for further experiments, to know the size (i.e. the number of constituents) and the total electric charge q of these entities delivered by the source. One of the Aun sources, including the ion optics and a Wien filter, is located at the Orsay MP Tandem terminal. A wide mass range of clusters corresponding to nq ratios of 5, 7, 9, 27, 40 (selected by the Wien filter) and even nanodroplets containing up to 400 atoms are produced, accelerated by the second stage of the accelerator and sorted out by the analysing magnet. According to the parameters set on the source and the Wien filter the selection of nq values is obtained with a distribution of charge and mass.
0168-583X/$ - see front matter 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.nimb.2004.06.005
580
S. Bouneau et al. / Nucl. Instr. and Meth. in Phys. Res. B 225 (2004) 579–589
In electrostatic accelerators the total particle energy depends on its charge. Therefore the measurements, after acceleration, of the cluster constituent numbers and velocities allow identification of the clusters (for a given tuning of the source). This is achieved by time-of flight measurements of the cluster atomic fragments, after passage of the incident beam through a thin Formvar foil and by measuring simultaneously, with a multianode detector [7], the number of constituents coming out of the foil. High energy clusters, accelerated at several MeV, are thus necessary to define and verify experimentally the working parameters of the cluster ion source device. The same type of source is now used to produce large size clusters with energies below 50 keV to investigate desorption phenomena with large projectiles as well as applications in mass spectrometry. For example, the secondary ion emission from bio-organic solids under the impact of a Au4þ 400 beam has been studied for the first time [8]. Several sets of source parameters have been experimentally explored and this work presents the method to characterize Aun cluster beams and the variety of beams which can be delivered by a gold LMIS.
2. Experimental process
selection of the ejected clusters is located at the accelerator terminal [10,13]. The first cluster beams were obtained, over a large energy range, more than ten years ago. To increase the cluster beam intensities injected into the accelerator, the diameter of the extractor aperture at the source exit is set at 1.5 mm which leads to a 75 mrad acceptance. m In this configuration, the mass separation Dm is between 15 and 20. With this device it is possible to obtain ions from Siþ to Auþ 1000 . Fig. 1 shows an example of mass spectrum for very heavy ions down to Auþ 5 . It shows the variation of the intensity measured by the Faraday cup, at the exit of the Wien filter, versus the voltage set on the filter plates. Clusters containing some tens of atoms were already obtained with this source [14,15], but in this spectrum a maximum of beam intensity is observed around a potential value of 10 V corresponding to mq 20 000. This ratio is deduced from the relation rffiffiffiffi q VWien þ d ¼ K I ; m VWien (V): voltage on the electrostatic deviation plates, I (A): current intensity in the coils generating the magnetic field, K: constant depending on the alignement of the source and Wien filter axes, experimentally adjusted.
2.1. Beam production 2.1.1. The ORION injector The gold clusters are provided by a liquid metal ion source, duplicate of the pattern developed by Sudraud and Ben Assayag [9] the specifications of which are described in [2,10,11]. This simple source is made of a tungsten needle and of a reservoir containing the metal. An electric field, applied between the needlepoint and an extractor, generates a Taylor cone from which are emitted ions, clusters and nanodroplets. These are further accelerated by a potential of some tens of kilovolts. To obtain a low melting point, 370 C, the reservoir contains an AuSi eutectic with 31 silicon atoms for 100 gold ones (i.e. a mass ratio of 4.4% silicon with respect to gold) [12]. The whole set-up including the ion source, two focusing lenses and a Wien filter for the mass
Fig. 1. Current delivered by the LMIS source versus the VWien voltage.
S. Bouneau et al. / Nucl. Instr. and Meth. in Phys. Res. B 225 (2004) 579–589
The offset value d depends on the beam angle at the filter inlet and a mass calibration of the Wien filter, with well known beams (Auþþ , Auþ , Auþ 2, Auþ 5 , . . .), is advisable for each new source configuration. Above nq ¼ 9 it is not possible any more to separate the cluster masses since the curve of Fig. 1 becomes continuous. However, one can modify the intensity of the cluster beam as well as the maximum of its mass distribution through the adjustment of three main parameters: the voltages on the focusing lenses, the current emitted by the source and the acceleration voltage.
581
i(ii) Fig. 3 shows the importance of the source emission current, as large size cluster beams, obtained for low values of the Wien filter potential (Fig. 1), have their highest intensity for I ¼ 100 lA. But a compromise must be found between the current value and the lifetime of the ion source (from 300 to 1000 h). (iii) The acceleration voltage value is, in this case, fixed at 10 kV. In the following, since the same nq ratio may be related to different cluster size and charge state values, two sets of source parameters are used for producing either low charged or highly charged clusters, which also corresponds to light or heavier ones.
ii(i) The voltages (U1 and U2 ) applied on the two focusing lenses allow to optimize the intensity of the selected clusters, compensating for the energy spread due to the ionization of clusters at some distance from the needle head. This energy spread increases with the cluster mass [15] and can reach 100 eV or more. As can be seen in Fig. 2, the focusing conditions play an important role for ‘‘light’’ clusters but are nearly the same from nq ¼ 9 up to 100. However, the focusing conditions can also change the shape of the beam intensity curve. n The Auþ n peaks, for q P 9, may disappear while a ‘‘background’’ increases which is not an actual background but the sum of the 3þ 2þ 3þ Au2þ 2n1 , Au3n1 , Au2nþ1 , Au3nþ1 ,. . . peaks.
2.1.2. Acceleration and beam line At the Tandem terminal the selected clusters are injected, using two electrostatic mirrors, in the high energy (HE) accelerating tube in order to reach energies from 1 to 12q MeV. The residual gas pressure in the tube and along the beam line is kept lower than 107 h Pa by ion and titanium vapor pumps. The beam line includes, at the machine exit, two electrostatic quadrupole lenses together with X and Y steerers, to focus and direct the beam to an analysing magnet followed by a diagnostic chamber. In the latter, by means of sliding slits associated with a microchannel plate
Fig. 2. Optimized values of the voltages (U1 , U2 ) applied on the two focusing lenses versus the n=q ratios of the selected clusters.
Fig. 3. Intensity of the cluster beam as a function of the Wien filter potential for different source emission currents.
582
S. Bouneau et al. / Nucl. Instr. and Meth. in Phys. Res. B 225 (2004) 579–589
detector (MCP) and a Faraday cup, the beam can be centered and collimated. For nq 6 40, the diagnostic and collision chambers are set at a 1.29 angle with respect to the beam axis and the analysing magnet eliminates the fragments of the projectile, mainly due to collisions with the residual gas. The magnetic field necessary to bend the heavier cluster trajectories would be too high and the measurements are then performed at 0 without rejection of the background due to the fragments.
detector records, for each incident projectile, the number of simultaneous impacts (multiplicity) on the detector. The data accumulation over a large number of events provides a multiplicity distribution with a centroid M. Knowing the detector efficiency (see Appendix A), the average number of constituents n ¼ M is obtained as well as the average charge q as nq ¼ nq. Furthermore, the time-of-flight measurement between the foil and the multianode detector gives the velocity of the Au atoms exiting the Formvar foil:
2.2. Method used to characterize the ion beams DT ¼ A schematic view of the experimental set-up is given in Fig. 4. The clusters with given nq values are selected by the Wien filter, pre-accelerated by a dU voltage in the LMIS source device and further accelerated from the terminal at a potential U to the collision chamber. The total energy for a charge q cluster is then E ¼ q½Uterminal þ dULMIS :
ð1Þ
For each charge q there is a distribution of n due to the limited mass separation power of the Wien filter. A variety of clusters, with different q values and constituent numbers, are thus accelerated. After passing through a Formvar foil perpendicular to the beam direction, a cluster dissociates into its atomic constituents and the pixelated
L L ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffi ; 2ðE1 DEÞ v mAu
where DE is the calculated energy loss in the thin foil, mAu the gold atomic mass and L the foildetector distance. The time-of-flight measurement of the atomic constituents allows verification of the correct assignment of the nq ratio, deduced from the Wien filter parameters. Experimentally a window set on the time-of-flight peak is then used to remove the background in the measured distribution of the constituent number. 2.3. Detection 2.3.1. Short description The electrons and negative ions emitted at the foil entrance are deflected at 90 by an electrostatic
Fig. 4. Schematic view of the experimental set-up.
S. Bouneau et al. / Nucl. Instr. and Meth. in Phys. Res. B 225 (2004) 579–589
mirror towards an MCP detector; the pulse delivered by the electron impact gives the start of the time-of-flight (Fig. 4). Furthermore, for nq P 40, the shape of the signals due to the electrons is visualized and recorded by a digital oscilloscope, to make sure that they actually correspond to a single impact and not to several successive fragments. It is also possible to choose the H ions as a start signal. The projectile constituents, after the foil, are collected by a multianode detector; two different detectors of this kind have been used. The first one, described in [7], is made of a set of two MCPs (Hamamatsu, / ¼ 42 mm, geometrical efficiency 0.6) and of a 256 pixel anode array (1600 mm2 ); it will be called D256 hereafter. The other smaller one (D64), has three MCPs (Hamamatsu, / ¼ 28 mm, geometrical efficiency 0.66) and a 64 pixel anode array (400 mm2 ). With these detectors, located along the beam axis at 256 mm from the target, it is possible to measure the number of constituents and their time-of-flight. The use of the SRIM code [16] to compute the angular distributions of the constituents coming out of the foil allows calculation of the proportion reaching the detector, listed in the last column of Table 1. Four formvar foils, with known thicknesses, are used in these experiments and the time-of-flight Table 1 Summary of experimental conditions n Energy/Au Deviation q (MeV) angle ()
583
calibration is performed with a pure Au5 beam at a known energy. 2.3.2. Detector efficiency A more complete treatment of this subject is reported in Appendix A. Two efficiencies are involved for the detector: 1 is obtained by multiplying, for each beam, the MCPs geometrical efficiency by the percentage of intercepted ions and 2 takes into account the pixelated structure of the anode array. The efficiency 2 is related to the geometrical structure of the anode and to the discriminator thresholds [7]. The current delivered by the set of MCPs, in these experiments, is always sufficient to detect the impact of gold atomic ions. The 2 value corresponds to the detection of an impact giving a single triggering in the pixels, whereas (1 2 ) is associated to the detection of an impact in the interpixel zone leading, in our case, to a simultaneous triggering of two adjacent pixels. The 2 value is close to the ratio 2 ’
active geometrical surface ¼ 0:74: total geometrical surface
We took this value as a starting point for the multiplicity analysis, given in detail in the next section, and 2 has been adjusted to get the best agreement with experimental results.
Formvar foil thickness (nm)
Detector
Ion source conditions
% of intercepted ions
5 7 9 27
2.008 1.434 1.115 0.372
1.29 – – –
49 ± 2 – – –
D64 / ¼ 28 mm – –
Low charge Mode – –
66 53 44 13
5 40
1.808 0.225
1.29 –
25 ± 2 –
D256 / ¼ 42 mm
Low charge Mode
100 58
35 120 400
0.287 0.084 0.025
0 0.03 0
15 ± 2 – –
D256 – –
Low charge Mode –
81 38 22
9 27 40
0.2 0.2 0.2
1.29 – –
26 ± 2 – –
D256 – –
High charge Mode –
40 – –
584
S. Bouneau et al. / Nucl. Instr. and Meth. in Phys. Res. B 225 (2004) 579–589
3. Experimental results and discussion The LMIS source delivers Aun Sip clusters with a p=n ratio of 0.1 to 0.2, estimated from the mass spectra. In this paper, we take into account the number of gold constituents of the cluster, neglecting the silicon contribution. Through this approximation only 2% of the cluster mass is, in fact, neglected, which has no visible effect on the time-of-flight spectra. The silicon atoms are scattered, after the foil, in a solid angle about five times wider than that for the gold atoms and less than 10% of these reach the detector. So, the number of gold constituents measured is overestimated by 2% at the most, because of the unknown silicon atoms number. The experimental conditions for the various beams are summarized in Table 1. Two sets of experiments are performed and compared: one for a given total energy of the projectiles and the other for the same projectile velocity. 3.1. Time-of-flight For a given energy per atom (the total energy and nq ratio being known) one can calculate the energy loss in the Formvar foil and deduce the time-of-flight values. The measured time-of-flights values, up to nq 400, agree with the calculated ones. Furthermore, the energy spread in the foil, as well as scattering phenomena, give rise to angular
and energy distributions of the exiting atoms which can be estimated with the SRIM code [16]. As a result, the calculated time-of-flight peak shapes are also shown to be in agreement with the experimental data. An example of calculated curve is given in Fig. þ 5 for the Auþ 5 and Au9 projectiles, at the same 10.04 MeV total energy. The adjustment of these curves, as well as that of the Auþ 7 one (not shown), is obtained through the convolution of the time-offlight distribution, deduced from the SRIM energy distribution of the atomic ions, with a Gaussian distribution (r ¼ 3:8 ns) which accounts for the time-of-flight resolving power and the nonhomogeneity of the Formvar foil. The parameters used in the three cases are the same. 3.2. Determination of the gold cluster masses Fig. 6 shows the multiplicity (i.e. the number of impacts) distributions measured, in coincidence with a time window set on the time of flight peak, without magnetic selection, for voltages on the Wien filter plates corresponding to nq ¼ 35, 120, 400. In Fig. 6, the mean and maximum values of the measured distribution of impact numbers increase with the nq selection of the Wien filter and it is possible to obtain a rough nq estimate. A peak of high intensity at low multiplicity, due to the previously mentioned fragmentations, disappears when the beam is selected by the H.E.
7000 +
Au5
1000
5000
Counts/channel
Counts/channel
6000
4000 3000 2000 1000
180
(a)
+
Au9
100
10
230 240 250 260 270 280 290 300 310 320 330
200
Time (ns)
fit Experimental data
(b)
Time (ns)
þ Fig. 5. Experimental and calculated time-of-flight spectra for the Auþ 5 (histogram ¼ experiment) and Au9 projectiles (the time-of-flight offset value is 89 ns in both cases).
S. Bouneau et al. / Nucl. Instr. and Meth. in Phys. Res. B 225 (2004) 579–589
585
6
10
5
n/q=120
4
10 Counts
Probability of detection
10
n/q=35 3
10
2
10
1
10
n/q=400
0.1
0.01 M1 M2 M3 M4 M5 Sum of all contributions experimental data
1E-3
0
10
0
100
200
300
400
500
Number of impacts Ni on the D256 multi-pixel detector
Fig. 6. Number of counts versus the measured multiplicity, without magnetic field and for several Wien filter settings, in the low mass ion source conditions.
analysing magnet. However, for nq P 40 it is possible to use the H signal, which emission yield exceeds 100%, as a start signal to reduce the contribution of the fragmentation in the spectra. The emission of H ions induced by fragments is, indeed, nearly negligible. 3.2.1. Influence of the multianode structure There are two stages for the calculation of the number of constituents corresponding to the detection of the clusters. The probability for the MCPs to detect p ions among n follows a binomial law function of 1 : P ðpÞ ¼ Cnp p1 ð1 1 Þ
np
:
The pixelated structure of the anode surface implies to take into account the 2 efficiency as defined before. The detection probability of all the anode central parts obeys also a binomial law, function of 2 , and the probability of detecting r ions among n is the product of these two binomials: P ðrÞ ¼ Cnr r1 r2 ð1 1 2 Þnr : An ion falling in the interpixel area may be detected by the two adjoining pixels. The binomial law is the same, but the corresponding multiplicity is M ¼ 2ðp rÞ. The variation of the detection probability in terms of the multiplicity finally assigns to each P ðrÞ value a M ¼ r þ 2ðp rÞ ¼ 2p r multiplic-
0
2
4
6
8
10
Multiplicity Fig. 7. Calculated contributions of the M ¼ 1–5 detected ions and comparison of their sum with the experimental distribution for Au5 .
ity. Then, the distribution of the detected ions as a function of the multiplicity reads, as before, XX P ðMÞ ¼ Cnp Cpr p1 ð1 1 Þnp r2 ð1 2 Þpr ; p
r
with M ¼ 2p r. The experimental multiplicity, obtained with the D256 detector, for a 9 MeV Auþ 5 beam, is displayed in Fig. 7. The measured multiplicity values go beyond 5 because of the multiple triggering of ions falling in the interpixel zone. Taking this into account as described above, an interpixel surface leading to 18% of double hits gives a perfect fit of the measured multiplicity spectrum. The events associated to one to five impacts (marked Mi , i ¼ 1–5) are detailed in Fig. 7. 3.2.2. Average mass measurement The various multiplicity curves, for the singlecharged clusters Au5 , Au7 , Au9 , were fitted with the already determined 1 efficiency (detection yield and solid angle calculation) and by adjusting the 2 parameter. An 2 value, lower than 0.74, indicates that the surface responsible for multiplicities of 2 is larger than the interpixel surface, this effect being related to the detector gain. For clusters with a nq ratio of 27 and 40, higher charge states must be considered to fit the distribution.
586
S. Bouneau et al. / Nucl. Instr. and Meth. in Phys. Res. B 225 (2004) 579–589
Table 2 Efficiencies of the two detectors for the studied clusters and measured average multiplicities M Source parameters
n=q
Low charge
5 7 9 27 40 120
High charge
D64
D256
E/atom (MeV)
1
2
M
E/atom (MeV)
1
2
2 1.43 1.11 0.37
0.42 0.28 0.24 0.08
0.57 0.6 0.73 0.7
3.46 3.4 3.38 9.04
1.8
0.60
0.82
0.225 0.084
0.35 0.23
0.8 0.8
0.2 0.2 0.2
0.24 – –
0.75 – –
9 27 40
In Table 2 are listed, for the different nq and energy values, the 1 and 2 efficiencies used and the measured average multiplicity; 1 and 2 have been adjusted, around the values provided by the experimental conditions, to reach the best agreement between measured and calculated distributions in terms of the multiplicity. The slight variation of 2 with the constituent number should be connected to the detector gain and to the influence of the velocity on the MCPs response under the impact of gold ions. 3.2.3. Charge state determination Fig. 8 spectra are obtained with the ion source parameters which favour the low mass cluster beams. From Fig. 8(a) one can see that, even in the
M 3.5
57.3 154 4.62 30.4 54.7
¼ 9 case, a small proportion of Auþþ 18 added to the prevailing Auþ contribution is necessary to repro9 duce the multiplicity distribution. In Fig. 8(b) are compared the experimental multiplicity distribution for nq ¼ 40 and the sum of the computed contributions of the different charge states involved. A good fit of the experimental result needs to include the q ¼ 1þ to 5þ charge states. The contributions of the various charge states which, when summing them up, give the best agreement between experiment and calculation for different nq values, are listed in Table 3, for two sets of the ion source parameters. The Table 3 values (percentages and deduced mean q value) are accurate within 10%. Table 3 shows that the relative contribution of the high charge states increases with nq, and that the
n q
Fig. 8. Comparison between the multiplicity distributions measured and computed for (a) nq ¼ 9 and (b) nq ¼ 40 taking into account several charge states.
S. Bouneau et al. / Nucl. Instr. and Meth. in Phys. Res. B 225 (2004) 579–589
587
Table 3 Percentages of the charge states involved for several nq values and for two ion source regimes n=q Low charge
High charge
q
q
1
2
5 7 9 27 40 120
98.7 94 96.6 24.6 1 3.1
1.3 6 3.4 18.8 11 7.4
3
4
19.2 36 16
15.5 44 20.9
13.8 7 18
13.4
9 27 40
23 4.1 2
74 13.3 4.5
3 26 12
34.8 25
16.6 31
6 19.5
ion source conditions play a crucial part. Going from the ‘‘low charge’’ to the ‘‘high charge’’ conditions, the average charge goes up by 80% for the ions with a nq ¼ 9 ratio, and by 35% for nq ¼ 40.
5
6
7
8
9
8.1 10.3
6
4.7
6
1.01 1.06 1.03 3 3.4 4.8 1.8 3.6 4.6
should now be available with a nanoparticle beam. Moreover, in the near future, the use of microbeams of these heavy clusters may be considered for imaging of biological samples, with 1–10 lm dimensions and on a 10 nm depth.
4. Conclusion Appendix A The LMIS source delivers gold clusters from qþ Auþ 2 to Au1000 or more. Previous results have shown that a cluster distribution exists for nq ratios between 10 and 1000 (i.e. masses between 2000 and 200 000 amu) with a maximum for nq ¼ 100 and with a beam intensity identical to that of Auþ 5. The cluster ions produced by a LMIS source and selected by a Wien filter are identified for a large qþ mass to charge range going from Auþ 5 to Au120q . The mass is deduced from the measurement of the constituent number of the cluster, accelerated at an energy of several MV q and broken up by passing through a thin foil. It is thus possible to obtain the mass and charge distributions of the cluster ions. For the maximum nq value explored here (120), the mean charge value is 5 and the mass distribution 9þ extends from Auþ 120 to Au1080 . A correlation has been established between the mass and charge distributions and the ion source working conditions. These experiments lead to a better knowledge of the LMIS source parameters and possibilities, in the fields of secondary emission and ion implantation in the keV energy range. Secondary ion emission rates greater than those ever observed
A.1. Featuring of the multianode detectors It is achieved following the procedure worked out in [7]. Three parameters must be determined: • The geometrical efficiency of the first MCP detector. • The detection loss due to the discriminator thresholds depending on the charge collected by the anodes, i.e. on the number of electrons emitted under ion impact and on the whole detector gain. • The efficiency of the anode array made of pixels with, in the interpixel zone, a separating grid surrounded by an insulating area to eliminate cross-talk between anodes. This configuration induces a special response of the interpixel. If the electron cloud coming out of the last MCP is in the interpixel zone, the respective positions of this cloud and of the interpixel center give rise to three possibilities: A single anode is hit and the incident ion is detected by one anode.
588
S. Bouneau et al. / Nucl. Instr. and Meth. in Phys. Res. B 225 (2004) 579–589
The two anodes are hit simultaneously and the ion is recorded twice. The charge collected by the anodes is not sufficient and the impact is not observed. The anode array efficiency splits up into the three probabilities:
Pdet : ¼
Ncoinc: N1B 1 ; N0 N0B D
ðA:6Þ
1 ¼ Pdet : D:
P0 no detection, P1 detection by a single anode, P2 detection by two adjoining anodes,
ðA:7Þ
If a transmitting grid T is on the beam path, Eqs. (A.6) and (A.7) become
with the relation P0 þ P1 þ P2 ¼ 1: ðA:1Þ The basic idea for the determination of the absolute efficiencies, or detection probabilities, is the coincidence between two detectors recording the same event. If the total number of particles is NTot: the first detector gives NRef: ¼ N0 ¼ NTot: :
Actually, account must be taken of the detector backgrounds, N0B and N1B , as well as of the correction for the data loss in the discriminators (D). The detection probability Pdet : is then written as
ðA:2Þ
Pdet : ¼
Ncoinc: N1B 1 ; N0 N0B DT
ðA:8Þ
1 ¼ Pdet : DT :
ðA:9Þ
If the detection solid angle does not cover the whole emission angle, it is necessary to add another term x describing the average collection percentage of the detector This term is equivalent to the transmission of a grid set on the incident ions path,
For the second detector, which must be calibrated,
1 ¼ Pdet : DT x:
N1 ¼ 1 NTot:
ðA:3Þ
In our detector case, with a pixelated anode array, two procedures are available:
Ncoinc: ¼ 1 NTot:
ðA:4Þ
Ncoinc: N0
ðA:5Þ
• either a small dimension beam is used to scan the detector, which provides its actual efficiency in each ðx; yÞ point (the coincidence number then corresponds to an event with at least one hit anode),
and the number of coincidences is
1 ¼
is the effective detection probability.
ðA:10Þ
(a) (b)
0.6
0.1
0.4 Pixel 3. Multiplicity 1 Pixel 4. multiplicity 1 Pixel 3. Multiplicity 2 Pixel 4. Multiplicity 2 Total
0.3 0.2
Efficiency
Efficiency
0.5
Total Pixel 3. multiplicity 1 Pixel 4. multiplicity 1 Pixel 3. multiplicity 2 Pixel 4. multiplicity 2
0.01
0.1 0.0 1.2
1E-3
1.4
1.6
Position X (mm)
1.8
10.0
15.0
20.0
30.0
40.0
50.0
60
Threshold (mV)
Fig. 9. D64 efficiency measured (a) versus the detector position, for a 30 mV threshold (b) versus the threshold in the middle of the interpixel zone.
S. Bouneau et al. / Nucl. Instr. and Meth. in Phys. Res. B 225 (2004) 579–589
1 ¼
1 S
Z Z ðx; yÞ dx dy;
ðA:11Þ
S
where S is the surface of the detector or of an anode array cell; • or the beam covers the whole detector and the effective efficiency takes into account the anode efficiency as for a grid crossing the ions path (Eqs. (A.8) and (A.9)). To find out Eq. (A.1) terms, a scan with a sharp beam tests the interpixel response. The yields associated to a single anode or to two neighbouring ones are measured in terms of the position, and the P1 and P2 probabilities of Eq. (A.1) are so determined, as well as P0 marking the lack of detection. Fig. 9(a) shows an example of these measurements; it presents the response of an area of the D64 detector to a 100 lm · 100 lm Auþ 5 beam, with a 30 mV discriminator threshold. It can be seen, on this figure, that in the interpixel center (1.4 mm position) the multiplicity 2 yield is maximum (two active anodes), greater than the value for a single hit anode. On the other hand, the total detection yield is 26% smaller, in this position, than that obtained in the middle of the adjoining pixel. We so have indeed access to P0 ðx; yÞ, P1 ðx; yÞ, P2 ðx; yÞ. Obviously, the P0 , P1 and P2 probabilities do vary with the charge per impact delivered by the detector, and/or with the discriminator threshold as shown in Fig. 9(b) where this analysis is made in the middle of the interpixel. In our case, we take into account P1 and P2 , while P0 is included in 1 . The notations used in this work are 2 ¼ P1
and
1 2 ¼ P2 :
For this study, the discriminator thresholds are set at 10 mV, so that the impacts producing two electrons at least are detected, and the detector gains are adjusted to obtain pulses with an average amplitude of about a hundred mV. The electron emission from a carbon foil following a gold atom impact is of about 10–20 electrons, with a factor of 2 between 300 keV and 2 MeV [18], and a full cluster impact follows a sublinear law in n0:6 , n being the constituent number [17,18]. With these conditions for the electron emission, it is possible to
589
neglect the 10 mV threshold as it has no effect. We were able to verify that D ’ 1 and that the interpixel loss is minimum, which agrees with P0 ’ 0.
References [1] M.G. Blain, S. Della-Negra, H. Joret, Y. Le Beyec, E.A. Schweikert, Phys. Rev. Lett. 63 (1989) 1625. [2] M. Benguerba, A. Brunelle, S. Della-Negra, J. Depauw, M. Joret, Y. Le Beyec, M.G. Blain, E.A. Schweikert, G. Ben Assayag, P. Sudraud, Nucl. Instr. and Meth. B 62 (1991) 8. [3] H.H. Andersen, A. Brunelle, S. Della-Negra, J. Depauw, D. Jacquet, Y. Le Beyec, J. Chaumont, H. Bernas, Phys. Rev. Lett. 80 (1998) 5433. [4] A. Brunelle, S. Della-Negra, J. Depauw, D. Jacquet, Y. Le Beyec, M. Pautrat, K. Baudin, H.H. Andersen, Phys. Rev. A 63 (2001) 022902. [5] S. Bouneau, A. Brunelle, S. Della-Negra, J. Depauw, D. Jacquet, Y. Le Beyec, M. Pautrat, M. Fallavier, J.C. Poizat, H.H. Andersen, Phys. Rev. B 65 (2002) 144106. [6] M. Fallavier, R. Kirsch, S.N. Morozov, J.C. Poizat, J.P. Thomas, N. Wehbe, Phys. Rev. B 68 (2003) 140102. [7] S. Bouneau, P. Cohen, S. Della-Negra, D. Jacquet, Y. Le Beyec, J. Le Bris, M. Pautrat, R. Sellem, Rev. Sci. Instr. 74 (2003) 57. [8] A. Tempez, J.A. Schultz, S. Della-Negra, J. Depauw, D. Jacquet, A. Novikov, Y. Le Beyec, M. Pautrat, M. Caroff, M. Ugarov, H. Bensaoula, M. Gonin, K. Fuhrer, A. Woods, Rapid Commun. Mass Spectrom. 18 (2004) 371. [9] G. Ben Assayag, Thesis, Universite Paul Sabatier, Toulouse, 1984. [10] P. Attal, S. Della-Negra, D. Gardes, J.D. Larson, Y. Le Beyec, R. Vienet-Legue, B. Waast, Nucl. Instr. and Meth. A 328 (1993) 293. [11] K. Boussofiane-Baudin, Thesis, Universite de Paris Sud, Orsay, 1993, IPNO-T-93-06. [12] Di Capua, Rend. Accad. Nazion. Lincei 29 (1920) 111. [13] B. Waast, S. Della-Negra, A. Lafoux, Nucl. Instr. and Meth. A 382 (1996) 348. [14] P. Sudraud, C. Colliex, J. van de Walle, J. Phys. Lett. 40 (1979) L207. [15] P.D. Prewett, G.L.R. Mair, in: P.S. Walsh (Ed.), Ion Beams from Liquid Metal Ion sources, Research Studies Press, 1991. [16] J.F. Ziegler, J.P. Biersack, U. Littmark, SRIM 2000.39, www.SRIM.org, in: J.F. Ziegler (Ed.), Stopping Powers and Ranges of Ions in Matter, Pergamon Press, New York, 1985. [17] M. Fallavier, Y. Champelovier, M. Ferrari, R. Kirsch, J.C. Poizat, J.P. Thomas, B. Canut, M. Monchanin, S.M.M. Ramos, P. Thevenard, S. Della-Negra, J.P. Mouffron, P. Nicol, Eur. Phys. J. D 9 (1999) 629. [18] K. Baudin, A. Brunelle, S. Della-Negra, J. Depauw, Y. Le Beyec, E.S. Parilis, Nucl. Instr. and Meth. B 117 (1996) 47.